Effect of a rigid wall on the vortex induced vibration of two staggered circular cylinders

Transcription

1 Effect of a rigid wall on the vortex induced vibration of two staggered circular cylinders Javad Farrokhi Derakhshandeh a), Maziar Arjomandi, Benjamin S. Cazzolato and Bassam Dally Abstract School of Mechanical Engineering University of Adelaide, Adelaide, South Australia 55, Australia Vortex Induced Vibrations (VIV) plays a key role in a wide range of engineering applications including the extraction of renewable energy. In this paper, numerical studies of the phenomenon of VIV were conducted to investigate the flow behaviour around two identical circular cylinders. The upstream cylinder was located in the vicinity of a rigid wall and downstream one was mounted on an elastic support with one degree of freedom (-DOF). The Reynolds number based on the cylinder s diameter was kept constant at 8,7, while the separation between the upstream cylinder and the wall was varied. The results show that this separation distance known as the gap ratio has a significant effect on the dynamic behaviour of the upstream and downstream cylinders. Accordingly, the interaction of shear layers between the upstream cylinder and the rigid wall has a strong influence on the vortex dynamics of both cylinders, in particular when the upstream cylinder was mounted close to the wall. In this arrangement, a jet flow produced in the wake of the upstream cylinder significantly affects the vortex shedding frequency, and the lift and drag coefficients of both cylinders. This can alter the dynamic response of the downstream cylinder and theoretical efficiency of the VIV power. Keywords: Vortex Induced Vibration, vortex shedding, gap ratio and hydrokinetic energy I. Introduction The conventional sources of energy, such as fossil fuels, are environmental pollutants and rapidly depleting. Therefore, it is necessary to seek alternative sources of energy, preferably those available freely in nature, such as renewable energies. The increasing need for renewable energy has motivated the harness of energy from the ocean and water resources such as currents and shallow rivers. Considering that almost two third of the earth s surface is covered with water and water resources are accessible, it is obvious that harnessing hydrokinetic energy can be a substantial source of renewable energy in the near. Guney and Kaygusuz [] in their report expected that hydropower energy production, including ocean power, provides up to 2 GW of installed production capacity by 225. Hence, hydropower energy can be recognised as an emerging source of electricity production. Recently, the Vortex Induced Vibration (VIV) converter has been invented to harness the kinetic energy of ocean currents to produce electricity [2]. In general, VIV is a destructive phenomenon and related to the response of bluff bodies immersed in a fluid flow. VIV can be characterised by both fluidic parameters such as dynamic forces, and structural parameters including the mass and damping ratios. Due to the interaction of these parameters, extensive studies have been dedicated to investigate the VIV response of a circular cylinder.

2 From the literature it has become clear that, VIV can be considered as a cost effective source of energy [3]. Although using hydropower as a source of energy has its own challenging, such as structural corrosion, employing a new technology to harness the abundance clean energy from currents and oceans with high density, low maintenance, sensible efficiency and long life is always desirable [4]. Among different hydropower sources of energy VIV appears the most advantageous in terms of production cost and volumetric power density. The comparison between conventional sources of energy and VIV energy generation indicates that the cost of VIV is comparable with those sources of energy [3]. In a benchmarking study, VIV energy convertor was compared to three types of wave energy convertors including Pelamis, the Opt power buoy and Energetech. Pelamis is a floating ocean wave convertor; Opt power buoy is another floating wave convertor which is moored to the sea bed, and Energetech, is located on the sea bed or moored at a low depth which has been developed in Australia in 992. Figure shows that the volume density of VIV (kw/m 3 ) is much higher than other types of wave convertors [3]. Figure : Comparison of volume energy density of VIV with three types of wave convertors; Based on Bernitsas et al. [3]. Vortex formation of an oscillating bluff body have been studied by Bearman [5]. This study modelled the behaviour of an elastically mounted cylinder as a simple mass-spring-damper. Khalak and Williamson [6] measured shear forces on a circular cylinder in a water channel. They classified the amplitude response of a circular cylinder for a very low mass and damping system into three categories; initial, upper and lower. In the initial category, the oscillation of the cylinder begins to build up and in the upper one the resonance occurs with the maximum obtainable amplitude ratio. Finally, at the lower amplitude response, the oscillation of the cylinder is damped. Govardhan and Williamson [7] showed that the maximum amplitude of oscillation is a function of reduced velocity which is defined as = ( ), where U is the free stream velocity, f is the frequency of oscillation and D is the diameter of the cylinder. They demonstrated that at reduced velocity between = 5 and, the upper category of oscillation (y/d = ) is achievable using the low product of mass and damping ratios ( =.). The mass and damping ratios can be defined as =4 and = 2 respectively, where m is the mass of the system, is the length of the cylinder, is the viscous damping coefficient, and is the spring stiffness. Scholars also have determined the efficiency of the VIV power both theoretically and experimentally for different scenarios. Bernitsas, Raghavan [3] invented a converter to capture the VIV energy of ocean currents, and developed a mathematical model to calculate the theoretical efficiency of VIV power. In their investigations, the experimental and 2

3 theoretical VIV efficiencies were estimated to be 22% and 37%, respectively. Furthermore, Raghavan and Bernitsas [8] showed that the Reynolds number has a significant effect on the amplitude ratio of the cylinder and play an important role than the mass and damping ratios. Their experiments in the Transition Shear Layer (TrSL3) regime (2 l 4 4 l 4 < Re < 3.5 l 5 6 l 6 ) was found to produce a maximum amplitude ratio of y/d =.8 with a mass damping ratio 2 times higher than that of Govardhan and Williamson [7]. It is shown that the maximum amplitude of the oscillation can considerably affect the efficiency of the VIV power. Farrokhi Derakhshandeh J. [9] investigated the effect of arrangements of two staggered circular cylinders on the efficiency of the VIV power at high Reynolds number. They modelled 35 arrangements of the identical cylinders based on longitudinal (x/d) and lateral (y/d) distances between cylinders. According to their studies, the best location for the downstream cylinder to achieve the maximum efficiency of VIV was 3.5 x/d 4.5, and y/d 2 with approximately 48% of theoretical efficiency. All previous studies were generally conducted to analyse the response of VIV for cylinders which were mounted without any blockage effects. Apart from the flow around unbounded cylinders, there are several studies of a flow over a single cylinder with different cross sections near the wall. The studies of flow around a single cylinder in the vicinity of a rigid wall have shown that the form of the dynamics of vortex formation differs noticeably with changes of the gap ratio (G/D), where G is the separation between the cylinder and the wall. The effect of a wall on the cylinder generates a non-uniform velocity profile in the wake of the cylinder. Consequently, asymmetric shear forces affect the cylinder, and a jet flow behind the cylinder is generated. This fact has been highlighted by previous researchers experimentally and numerically [-3]. In these studies, the critical gap ratio (G/D) in which the suppression of vortex shedding occur, was found to be approximately in the range of.3 G/D.45. Although the previous studies have highlighted the influence of different parameters on VIV for unbounded and bounded cylinders, the authors are not aware of any investigation which analyse the flow around bounded cylinders to extract VIV energy. Hence, in the current paper, the response of the elastically mounted downstream cylinder in the wake of the stationary upstream cylinder is considered. In this staggered arrangement, the gap ratio of the upstream cylinder differs from.25 to.5. This study aims to calculate the theoretical efficiency of the VIV energy that can be harnessed by a downstream elastically mounted cylinder, as a function of the gap ratio of the upstream cylinder. II. Methods In this paper, Computational Fluid Dynamics (CFD) analysis was implemented using ANSYS Fluent package to determine the shear forces, the maximum amplitude of the downstream cylinder and the phase lag between the force and displacement of the downstream cylinder, which is necessary to determine the amount of power generated by the device. The displacement of the downstream cylinder was modelled using the dynamic mesh of ANSYS Fluent. Of the three types of dynamic mesh methods (namely smoothing, layering or 3

4 remeshing) available in Fluent, the remeshing method is the most suitable for unstructured meshes and the layering approach is employed for the case where the entire mesh moves up and down. The diffusion smoothing is the most suitable for quadrilateral mesh types employed in this study. By setting up the diffusion method as a boundary distance, the diffusion mainly influences the interior cells and to a lesser extent the cells adjacent to the moving wall. This approach assists in achieving a consistent y-plus (non-dimensional wall distance for a wall-bounded flow) value, regardless of the motion at each time step. To study the characteristics of the flow around the cylinders, the Scale Adaptive Simulation (SAS) turbulence model of ANSYS Fluent was used based on the Finite Volume Method (FVM) with a pressure based algorithm. In comparison to the other turbulence models, the SAS model is a relatively new model and was developed by Menter and Egorov [4]. This model uses the von Karman length scale which is defined as: = / /. () Here, is the turbulent kinetic energy. Using the von Karman length scale enables the model to adapt its behaviour to Scale Resolving Simulation (SRS) according to the stability parameters of the flow [5]. This allows the model to provide a balance between the contributions of the simulated and resolved parts of the turbulence stresses. Hence, the model can effectively and automatically switch from the Large Eddy Simulation (LES) mode to the Reynolds Average Navier Stokes (RANS) mode [6]. Although the SAS model is not as accurate as the LES model, it is less dependent on the mesh element resolution compared with the LES models and can capture the turbulence length scales more accurately than k ω, k ε or even Shear Stress Transport (SST) models [7]. Furthermore, Guilmineau and Queutey [8] showed that SST model cannot predict the maximum amplitude of oscillation in the upper category of oscillation. Therefore, compared to the LES model, the SAS model can be utilized with a coarser mesh while still keeping its accuracy. This advantage of the SAS model reduces the computational cost without compromising the model accuracy. A detailed explanation of the SAS model has been reported elsewhere [4, 6, 7]. A. Mathematical model The response of an elastically mounted cylinder with -DOF (here, the degree of freedom is in y-direction perpendicular to the free stream velocity) can be approximated by a massspring-damper model (Bearman 984). Therefore, the equation of motion can be written as!"(#)+!%(#)+!(#)=& (#), (2) where!% and!" are the velocity and acceleration of the cylinder respectively, and & is the force exerted on the cylinder by the fluid perpendicular to the direction of flow. In the VIV phenomenon, the natural frequency of oscillation is considerably influenced by the additional mass of fluid that is accelerated with the bluff body. For instance, the added mass of bluff body immersed in air is often negligible, however a bluff body immersed in water can have significant additional mass. For this reason, the mass ratio is defined as non- 4

5 dimensional parameters based on the specific mass of the fluid in which the cylinder is immersed [9]. Therefore, the structural parameters including the mass ratio and structural damping are generally expressed by =4 and = 2, respectively [2]. Here, presents the fluid density. By assuming linear behaviour and sinusoidal response of the cylinder, the time dependent fluctuation amplitude, the velocity and lift coefficient of the cylinder can be obtained from the following equations respectively,!(#)=! ()* sin(2. #), (3)!%(#)= (2. )! ()* cos(2. #), (4) (#)=2.sin(2. #+4), (5) where! ()* is the maximum amplitude,. represents the frequency at which vortices shed,!% and are the time dependent velocity and lift coefficient of the cylinder, 2 is the lift coefficient amplitude, and 4 is the phase angle of the displacement with respect to the exciting lift force, which for a linear system at resonance is close to /2 [3]. The total work produced by the downstream cylinder may be obtained using the vertical force, comprising the pressure and viscous forces on the cylinder. Hence, the work acting on the downstream cylinder can be calculated as the inner product of the total vertical force by the displacement of the cylinder in a one complete cycle of oscillation (5). ; =9 &!%:#. (6) By integrating the right hand side of Equation (6), the power of VIV can be written as follow: < = 787 = >?@? ;, (7) = 787 = A π 2.! ()* sin(4). (8) The detailed derivation of Equation (8) can be found in Bernitsas, Raghavan [3]. Finally, the fluid power can be used to calculate the efficiency of the VIV power over a complete cycle as follows: C 787 = = 787 = DEFGH, (9) where = DEFGH is the fluid power based on the dynamic pressure which is extracted from the Bernoulli s equation A. Therefore, correlating the wetted area of the cylinder, DL, and dynamic pressure results in the force over the projected area. Hence, the power in the fluid can be written as = DEFGH =.5 I and the efficiency of VIV can be defined as: 5

6 C 787 = A π 2.! ()* sin(4) A I which in terms of the Strouhal number, St, is = π2.! ()* sin(4) C 787 =π 2 J! ()* KStsin(4). () Equation () shows that the theoretical efficiency of VIV power depends on four nondimensional parameters namely; the lift coefficient, the maximum amplitude of oscillation normalized by the cylinder diameter, the Strouhal number (St), and the phase angle between the exerted lift force on the cylinder and the displacement of the cylinder. In the current study, the impact of the key parameters of Equation () on the VIV efficiency is analysed. B. Problem definition A two-dimensional schematic of two staggered cylinders is shown in Figure 2. In this arrangement, the diameters of the cylinders were kept equal to D = 5mm. To facilitate the comparison, the Reynolds numbers and structural parameters were chosen based on the previous studies by Yang, Gao [2]. The Reynolds number based on the diameter of the cylinders, was kept constant at Re = 8,7. This Reynolds number corresponds to U =.74m/s. The structural parameters including the mass and damping ratios were set at = 3.87 and =.52, respectively. Therefore, the mass of the cylinder was defined in the User Define Function (UDF) file as m =.97kg. In addition, the natural frequency of the structure was set at f =.27 Hz. The fluid density was constant and close to the density of water at kg/m 3. The upstream cylinder was located at D from the inlet boundary. The turbulent boundary layer thickness over a flat plate can be estimated by N.382 x/ Re A/R [22]. For the chosen Reynolds number at this location, the thickness of the boundary layer was approximately N = 3.mm or N/ =.622. In the numerical model, the gap ratio between the upstream cylinder and the wall was varied from.25 to.5 with an increment of.25, which means that for Test Cases and 2, the upstream cylinder was partially submerged in the boundary layer (Table I - shaded area). For instance, at G/D =.25, the distance between the wall and the upstream cylinder is G = 2.5mm which means the upstream cylinder was submerged in the boundary layer by 8.6mm. Due to the blockage and tripping effects, the downstream cylinder was found to be affected by the jet and wake of the upstream cylinder. Six gap ratios of the upstream cylinder and two longitudinal locations, (H/D), of the downstream cylinders (in total twelve arrangements) were analysed. By increasing the gap ratio from G/D =.25 to.5, the angle between the cylinders (S) was reduced from 42.5 to 8.4 (Figure 2-b). The effect of the position of the downstream cylinder on VIV response is analysed using two longitudinal locations. Table I summarises the geometrical dimensions of the twelve test cases modelled in this study. 6

7 Table I: Geometrical dimensions of twelve test cases including the gap ratios, G/D, H/D and angle between the cylinders (S). Test Case * 2 * G/D H/D S(deg) * First cylinder is partially submerged in the boundary layer Considering these features, two main case studies were defined and analysed based on the structural support of the downstream cylinder. Case Study I includes twelve test cases in which both cylinders were mounted on the stationary supports to investigate the fluid flow (Table I). This assists to analyse the key parameters of the flow, such as shear forces and the frequency of vortices. Case Study II considers the effect of the stationary upstream cylinder on the downstream elastically mounted cylinder, with the same geometrical parameters are summarised in Table I, to analyse the efficiency of the VIV power. The quadrilateral mesh element was utilized for all numerical models. The mesh created was approximately 45, elements. Farrokhi Derakhshandeh et al. [9] previously showed that this number of mesh elements is sufficient to accurately capture the flow pattern around two cylinders in a turbulent flow. Considering the minimum mesh size and free stream velocity, the selected time step was set at.5s to achieve a converging solution. A typical structured mesh is shown in Figure 3.. III. Results Numerical simulations first were validated using experimental studies conducted on flow around a single stationary cylinder under the effect of a rigid wall [23, 24]. Following the validation, the results of numerical analyses for both case studies (Case Studies I and II) are presented in this section. 7

8 a) b) Figure 2: a) Schematic of the computational domain including the cylinders arrangement and boundary conditions, b) Schematic of the arrangements of cylinders; 6 positions for the upstream cylinder and 2 for the downstream cylinder. Figure 3: Typical mesh grid around two circular cylinders for the Test Case 3 (see Table I). A. Validation of the model For the ease of comparison, the Reynolds numbers were chosen similar to the ones used in the experiments of Price, Sumner [23] and Khabbouchi, Guellouz [24] (Re =,4, 3,9 and 8,7). Typical results of dimensionless mean velocity ( UV W ) at G/D =. are compared with experimental data of Khabbouchi, Guellouz [24] at six sections behind the cylinder (Figure 4). Here, UX is the average of the x-velocity, which was calculated over the total running time. The trend of the velocities curve can be seen in the wake of the cylinder at different sections. The results show that the behaviour of the velocity profile and size of the recirculation region of the numerical findings well matches to the recorded experimental results. However, some differences occur at x/d = and x/d =.5 close to the cylinder where the numerical results show a V-shape profile compared to the U-shape of experimental results. The V-shape profiles of the numerical findings is due to the finer meshes around the cylinder where the meshes have been created by bias factor compared to the coarse meshes far from the cylinder (see Figure 3). Therefore, the mean streamwise velocity profile of finer 8

9 meshes retains the V shape at lower wake sections. It is obvious that both characteristics of the size and form of the recirculation region is directly related to the length of the vortex formation region [25]. 5 u/u at G/D =. Exp. Khabbouchi et al.(22) y/d x/d Figure 4: Comparison of the mean streamwise ( UX W ) profiles in the near wake of the cylinder at gap ratio G/D =. and Re = 8,7 for a single cylinder. The experimental results extracted from Khabbouchi et al. [24] and are shown as red squares. The simulations are the blue circles. The Strouhal number is an important parameter in determining the VIV efficiency (Equation ) which has been obtained from the numerical simulations and is plotted in Figure 5-a as a function of the gap ratio along with the experimental results of Price, Sumner [23]. This figure shows that when the cylinder was positioned close to the wall, the Strouhal number increases due to the suppression of vortices between the cylinder and the wall. This is primarily because of the jet flow behaviour experienced by the generated vortices as they are quickly expunged from the cylinder vicinity. When the cylinder was located very close to the wall, G/D =.25, the Strouhal frequency profile shows two peaks at different ranges of Reynolds numbers. Figure 5-b demonstrates these two peaks at Re =,4. These two peaks occur due to the interaction of the boundary layers between the cylinder and wall which can lead to non-linearities. 9

10 a) b) Re=4 Re=39 Re=87 Price et al.22, Re=9 Price et al.22, Re= Re=4 St.3.25 PSD G/D St Figure 5: Strouhal frequency of a single cylinder; a) comparison of the Strouhal numbers at different gap ratios with the experimental data of Price et al. [23], b) power spectral density of the lift coefficient at G/D =.25 and Re =,4. B. Case study I Figures 6 and 7 depict the mean drag and fluctuating lift coefficients of the upstream and downstream cylinders, respectively. When the upstream cylinder is located in the vicinity of the wall, larger shear forces are produced due to the boundary layer effects. This increases the mean drag and fluctuating lift coefficients of the upstream cylinder. It can be seen that when the cylinder is located under the boundary layer, the drag coefficient increases at G/D =.5. At a lower gap ratio, G/D =.25, the upstream drag and fluctuating lift coefficients are lower due to the suppression of the vortices (Figures 6-a, and 7-a). Increasing the gap ratio to G/D.75, has minor impact on the drag and fluctuating lift coefficients. The sensitivity of these coefficients to the longitudinal distance of the downstream cylinder (H/D) is insignificant on both cylinders. Hence, when considering the VIV power of the downstream cylinder, it is deemed more appropriate to analyse only one group of the Test Cases where the H/D is equal to 3.5. The mean drag and fluctuating lift coefficients of the downstream cylinder are shown in Figures 6-b, and 7-b as a function of the gap ratio. It is observed that the mean drag coefficient of the downstream cylinder shows an opposite trend compared to the mean drag coefficient of the upstream cylinder at low gap ratios. However, the variation of the fluctuating lift coefficient of the downstream cylinder is limited to a narrow range approximated to be between.6 and.7. Since the downstream cylinder is placed in the outside of the boundary layer and can be only influenced by the jet flow of the produced vortices of the upstream cylinder, the drag coefficient of the downstream cylinder decreases to.5 at G/D =.5. The trend of the fluctuating lift coefficient of the upstream cylinder reveals that this parameter is more dependent and sensitive to the boundary layer (Figure 7- a). It can be seen that the variation of the fluctuating lift coefficient is between.3 and.8 at lower gap ratios and by increasing the gap ratio to G/D =.75, this factor decreases noticeably. This reduction can also be seen for the downstream cylinder.

11 a) b) C D.6 C D CD upstream,h/d=3.5 CD upstream,h/d= G/D.3 CD downstream,h/d=3.5 CD downstream,h/d= G/D Figure 6: Effect of the gap ratio and location of the downstream cylinder on the mean drag coefficients of the a) upstream cylinder and b) downstream cylinder at Re=8,7. a) b) C L.6 C L CL upstream,h/d=3.5 CL upstream,h/d= G/D.3 CL downstream,h/d=3.5 CL downstream,h/d= G/D Figure 7: Effect of the gap ratio and location of the downstream cylinder on the fluctuating lift coefficients of the a) upstream cylinder and b) downstream cylinder at Re=8,7. As mentioned earlier in Equation (), the lift coefficient is one of the parameters that can affect the extracted power of the VIV efficiency. However, Figure 7-b illustrates that the range of variation of the lift coefficient with respect to the gap ratio of the upstream cylinder is relatively small. C. Dynamics of Vorticity To analyse the effect of wall and jet flow, in the wake of the upstream cylinder, the typical profiles of the mean streamwise velocity for two gap ratios, G/D =.25 and., are depicted

12 in Figure 8. The data were normalized against the free stream velocity and the diameter of the cylinder. The streamwise velocity profile is shown in blue circles at X/D = 3 at which the downstream cylinder was positioned. It is observed that due to the wall effect, the mean streamwise velocity profiles are not symmetric and tend to be inclined towards one side. In addition, comparison between the streamwise velocity profiles at the same sections reveals that at lower gap ratio, G/D =.25, the streamwise velocity is higher than at the larger gap ratio G/D =., which means a larger jet flow in the gap is generated. This interaction alters the pressure coefficient of the downstream cylinder and it is expected to change the VIV response of the cylinder when the second cylinder was mounted elastically. 7 u/u at G/D = y/d x/d u/u at G/D = y/d x/d Figure 8: Mean streamwise ( UX W ) profiles in the near wake of the two cylinders at gap ratios G/D =.25 and, at Re = 8,7. Blue circles depict the mean streamwise velocity at x/d = 3 which is the position of the second cylinder. Vorticity contours of the Test Cases 2 and 5 over.8 seconds after convergence are shown in Figures 9 and, respectively. These test cases have been selected due to the critical values of the shear forces. The time period is sufficiently large to capture a complete cycle of vortex formation. 2

13 The vortices generated from upper and lower surface of the upstream cylinder have been identified in Figure 9. The positive vortices are generated from the lower surface while the negative ones arise from the upper surface. The pattern of vortex shedding in the wake of the upstream cylinder is referred to as two single vortex-modes, 2S, and indicating two single vortices in the wake of the upstream cylinder generated per cycle [26]. The presence of the rigid wall forces the positive vortex to move forward. (Figure 9, at t = 7.6s). The motion direction of the generated vortex depends on the gap ratio of the upstream cylinder. Figure 9 (G/D =.5) shows that these two vortices grow gradually and the negative vortex eventually rolls up over the positive vortex and covers the lower vortex. For the 2S mode, the transverse amplitude of vortices is small and it increases very slowly with increasing the Reynolds number or reduced velocity [27]. Consequently, during a complete cycle of the vortex formation, the produced upstream vortices pass the downstream cylinder without any contact and the space of S is generated between the vortices and downstream cylinder (see Figures 9 and, at t = 7.6s). The reason for this behaviour is related to the transverse amplitude of the single vortex mode. By increasing the gap ratio to G/D =.25, the positive vortices can still translate upward from the wall and S space gradually decreases to a minimum and finally at t = 7.6s the positive vortex connects to the lower surface of the downstream cylinder (Figure, t = 7.6s). The brackets around the vortices show that the maximum transverse amplitude of 2S vortex mode. Govardhan and Williamson [26] showed that at low mass damping ratios, three types of behaviour of VIV response occur. These are known as initial, upper and lower branches, in which the 2S mode corresponds to the initial branch of VIV response. With formation of the 2S mode in the initial branch, the intense strain region occurs outside the developing vortices; therefore, the vortices cannot be split and consequently, a high strain rate region can exist between the two generated vortices [26]. With 2S vortex mode, the maximum transverse force can arise from the initial to the upper transmission branches due to the effect of phase between force and displacement of the cylinder. Therefore, to enhance the influence of the positive vortices with limited transverse amplitude, not only the position of the downstream cylinder is important but also the forcedisplacement phase angle plays a vital role in harnessing the energy of vortices. Equation () demonstrates that the phase angle between force and displacement can considerably affect the generated work by the downstream cylinder. It is assumed that for a linear system at resonance, the phase angle between force and displacement for unbounded cylinders is /2 [3]; however, in reality and particularly due to the effect of the wall, the phase angle can change and affect the maximum amplitude of oscillation (! ()* ). In reference to Equation (), the maximum amplitude of oscillation is also an important factor which can directly influence the efficiency of VIV power. Hence, in the next section the importance of the lift coefficient, displacement amplitude and force-displacement phase lag on efficiency is studied. 3

14 Boundary layer thickness Bubble in the wake of upstream cylinder t=7.2s Negative vortex Transverse Amplitude of 2S vortices t=7.4s Positive vortex t=7.6s Direction of movement of produced vortex Space = S t=7.8s t=8.s 2S vortex mode in the wake of the upstream cylinder Maximum transverse amplitude of 2S vortices Figure 9: Instantaneous vorticity contours (s - ) at G/D =.5, H/D = 3.5, and Re = 8,7. 4

15 t=7.2s 2S vortex mode in the wake of the upstream cylinder t=7.4s t=7.6s Attack of positive vortex to the downstream cylinder Space = S Maximum transverse amplitude of 2S vortices t=7.8s t=8. s Figure : Instantaneous vorticity contours (s - ) at G/D =.25, H/D = 3.5, and Re = 8,7. 5

16 D. Case study II The efficiency of the VIV power for each test case can be calculated based on the obtainable parameters from the CFD model and using Equation (). The typical response of the downstream elastically supported cylinder under the produced jet flow of the upstream stationary cylinder is plotted in Figure. Included in this plot are the time series of the lift coefficient, the velocity, and the displacement of the cylinder (Test Case 2). Based on the maximum amplitude and the phase angle between force and displacement, the VIV response of all test cases as a function of the gap ratio has been summarized in Figure 2. The variation of the lift coefficient demonstrates that when the upstream cylinder is submerged in the boundary layer, the maximum lift coefficient of the downstream cylinder is achieved. Despite the maximum lift coefficient of the downstream cylinder at lower gap ratios, G/D.5, the amplitude of oscillations, y/d, remain at minimum. Outside of the critical gap ratio,.5 G/D., the amplitude of oscillation almost doubles and reaches a maximum. The reason for this is related to the phase angle between the lift force and the displacement of the downstream cylinder. It can be seen that the optimum phase angle is achieved at G/D = in which 4 = 9. It is observed that the phase angle can significantly affect the total generated work of the downstream cylinder. Hence, the force-displacement phase angle plays a more significant role in reducing the efficiency of the VIV power. To compare the efficiency of the VIV power from two cylinders under the effect of the boundary layer in vicinity of the wall, additional calculations were performed to analyse the efficiency of the VIV power for a single flexible mounted cylinder in a free stream flow. The numerical modelling was conducted with the same conditions as the bounded cylinders. Therefore, the fluid and structural parameters such as the Reynolds number and the density of fluid, the mass and damping ratios remains unchanged. C L Vel (m/s) y/d Time (s) Figure : Time history of the lift coefficient, velocity and displacement of the downstream cylinder at G/D =.5, Test Case 2. 6

17 Time series of the lift coefficient, velocity and displacement of the cylinder are plotted in Figure 3 and the results have been added in the last row of Table II. The numerical outcomes of a single circular cylinder, such as the lift coefficient, and the displacement of the cylinder are in reasonable agreement with those of the experiments (Assi 29). The numerical data show that for the unbounded cylinder, the efficiency of VIV power is limited to.% which is much lower than the efficiency of the bounded cylinders. C L Y/D St.2 sin(φ) G/D Figure 2: Non-dimensional displacement, the lift coefficient and force-displacement phase shift of the downstream cylinder as a function of the gap ratio of the upstream cylinder at Re = 8,7. C L Vel (m/s) y/d Time (s) Figure 3: VIV response of a single unbounded cylinder including time history of the lift coefficient, the velocity and the displacement of the cylinder at Re =8,7. 7

18 As shown in Equation (), the lift force and the displacement of the cylinder as well as the Strouhal number can alter the total efficiency of VIV power. The effect of all key parameters is summarised and plotted in Figure 4 as the efficiency of the VIV power. It is clear from this figure that the efficiency increases with an increase in the gap ratio, from G/D.25 to G/D. Consequently, the maximum efficiency of the VIV power among the test cases is found at the gap ratio of G/D = (C YZY = 37.2%). This efficiency is considerably higher compared to the efficiency of VIV power for an unbounded single cylinder. In addition, in comparison to a turbine based system the overall efficiency is comparable. A developed numerical model reported by Li and Calisal [28] and Li and Calisal [29] have shown an advantage of mounting a twin turbine system with an increase of 25% over the two standalone turbines. Further development of these systems will better enhance their power output and competitiveness with our energy convertors. Table II: Effect of gap ratio on the frequency of oscillation, the Strouhal number, the displacement, the velocity of the downstream cylinder, and the efficiency of VIV power G/D f (Hz) St y/d 2 sin( ) C 787 % Unbounded cylinder (Num. Results) Unbounded cylinder * (Exp. Results) * Calculation of VIV efficiency based on the fluidic parameters of Assi [9] at Re = 8, η (%) G/D Figure 4: Efficiency of VIV power as a function of the gap ratio of the upstream cylinder. 8

19 V. Conclusions Numerical analyses have been conducted to investigate the VIV response of two staggered circular cylinders using SAS turbulence model. The downstream elastically mounted circular cylinder was affected by the jet flow of the upstream stationary cylinder mounted next to a rigid wall. Fluidic parameters such as shear forces and the vortex shedding frequency of the cylinders were analysed. It was shown that the arrangements of the cylinders based on the gap ratio of the upstream cylinder can affect the lift and drag coefficients. Furthermore, the VIV response of the downstream cylinder was influenced by the jet flow of the upstream cylinder. At gap ratios.25 G/D.5 the drag and fluctuating lift coefficients change dramatically due to the effects of the boundary layer. Furthermore, key parameters such as; the lift coefficient, the maximum amplitude of oscillation, the Strouhal number and the phase angle between the exerted lift force on the cylinder and the displacement of the cylinder on the efficiency of the VIV power were analysed. Despite the maximum lift coefficient of downstream cylinder at lower gap ratios, the maximum oscillation of the downstream cylinder occurs at higher gap ratios (G/D ). This is due to the phase angle between the exerted lift force and displacement of the cylinder. It was observed that the maximum achievable theoretical efficiency of VIV power for the current test cases is approximately 37% at G/D = and it is much higher than the VIV efficiency of a single unbounded cylinder. Such findings have important implications on harnessing hydropower energy from VIV phenomenon. References [] Güney, M. and Kaygusuz, K., 2. Hydrokinetic energy conversion systems: A technology status review. Renewable and Sustainable Energy Reviews, 4 (9), [2] Bernitsas, M. and Raghavan, K., Converter of current/tide/wave energy. Provisional Patent Application. United States Patent and Trademark Office Serial, Serial No. 6/628,252,(24). [3] Bernitsas, M.M., Raghavan, K., Ben-Simon, Y. and Garcia, E., VIVACE (Vortex Induced Vibration Aquatic Clean Energy): A new concept in generation of clean and renewable energy from fluid flow. Journal of Offshore Mechanics and Arctic Engineering, Vol.3, - 5, (28). [4] Bernitsas, M.M., Ben-Simon, Y., Raghavan, K. and Garcia, E., The vivace converter: Model tests at high damping and reynolds number around. Journal of Offshore Mechanics and Arctic Engineering, Vol.3, -2, (29). [5] Bearman, P.W., Vortex shedding from oscillating bluff bodies. Fluid Mechanics, 6, , (984). [6] Khalak, A. and Williamson, C.H.K., Fluid forces and dynamics of a hydroelastic structure with very low mass and damping. Journal of Fluids and Structures, (8), , (997). [7] Govardhan, R. and Williamson, C., Critical mass in vortex-induced vibration of a cylinder. Eur. J. Mech.-B/ Fluids, 23 (), 7-27, (24). 9

20 [8] Raghavan, K. and Bernitsas, M.M., Experimental investigation of reynolds number effect on vortex induced vibration of rigid circular cylinder on elastic supports. Ocean Engineering, 38 (5-6), 79-73, (2). [9] Farrokhi Derakhshandeh J., Arjomandi., M., Dally, B. and Cazzolato, B., The effect of arrangements of two circular cylinders on the maximum efficiency of Vortex-Induced Vibration power using a Scale-Adaptive Simulation model, Journal of Fluids and Structures, (unpublished). [] Bearman, P. and Zdravkovich, M., Flow around a circular cylinder near a plane boundary. Journal of Fluid Mechanics, 89 (), 33-47,( 978). [] Taniguchi, S. and Miyakoshi, K., Fluctuating fluid forces acting on a circular cylinder and interference with a plane wall. Experiments in Fluids, 9 (4), 97-24, (99). [2] Mahir, N. and Rockwell, D., Vortex formation from a forced system of two cylinders. Part i: Tandem arrangement. Journal of Fluids and Structures, (5), , (996). [3] Sarkar, S. and Sarkar, S., Vortex dynamics of a cylinder wake in proximity to a wall. Journal of Fluids and Structures, 26 (), 9-4, (2). [4] Menter, F. and Egorov, Y., A scale-adaptive simulation model using two-equation models. In Proceedings of the American Institute of Aeronautics and Astronautics, -3, (25). [5] Egorov, Y., Menter, F., Lechner, R. and Cokljat, D., The scale-adaptive simulation method for unsteady turbulent flow predictions. Part 2: Application to complex flows. Flow, Turbulence and Combustion, 85 (), 39-65, (2). [6] Menter, F. and Egorov, Y., The scale-adaptive simulation method for unsteady turbulent flow predictions. Part : Theory and model description. Flow, Turbulence and Combustion, 85 (), 3-38, (2). [7] Menter, F., Garbaruk, A., Smirnov, P., Cokljat, D. & Mathey, F., Scale-adaptive simulation with artificial forcing. Progress in Hybrid RANS-LES Modelling, , (2). [8] Guilmineau, E. and Queutey, P., Numerical simulation of vortex-induced vibration of a circular cylinder with low mass-damping in a turbulent flow. Journal of Fluids and Structures, 9 (4), , (24). [9] Assi, G., Mechanisms for flow-induced vibration of interfering bluff bodies. PhD thesis, Imperial College London, London, UK, (29). [2] Khalak, A. and Williamson, C., Dynamics of a hydroelastic cylinder with very low mass and damping. Journal of Fluids and Structures, (5), , (996). [2] Yang, B., Gao, F., Jeng, D.-S. and Wu, Y., Experimental study of vortex-induced vibrations of a cylinder near a rigid plane boundary in steady flow. Acta Mechanica Sinica, 25 (),

21 [22] Schlichting, H. and Gersten, K., Boundary-layer theory: Springer Verlag, (2). [23] Price, S., Sumner, D., Smith, J., Leong, K. & Paidoussis, M., Flow visualization around a circular cylinder near to a plane wall. Journal of Fluids and Structures, 6 (2), 75-9, (22). [24] Khabbouchi, I., Guellouz, M. and Ben Nasrallah, S., A study of the effect of the jet-like flow on the near wake behind a circular cylinder close to a plane wall. Experimental Thermal and Fluid Science, 44, 285-3, (22). [25] Kravchenko, A.G. and Moin, P., Numerical studies of flow over a circular cylinder at Re = 39. Physics of Fluids, 2, 43-47, (2). [26] Govardhan, R. and Williamson, C., Modes of vortex formation and frequency response of a freely vibrating cylinder. Journal of Fluid Mechanics, 42, 85-3, (2). [27] Zhao, M. and Cheng, L., Numerical simulation of two-degree-of-freedom vortex induced vibration of a circular cylinder close to a plane boundary. Journal of Fluids and Structures, 27 (7), 97-, (2). [28] Li, Y. and Calışal, S.M., Modeling of twin-turbine systems with vertical axis tidal current turbines: Part I power output. Ocean Engineering, 37 (7), , (2). [29] Li, Y. and Calisal, S.M., Modeling of twin-turbine systems with vertical axis tidal current turbine: Part II torque fluctuation. Ocean Engineering, 38 (4), ,(2). 2